Frequency-domain techniques for heart rate variability analysis

Abstract

Two frequency domain techniques, fast Fourier transform (FFT) and maximum entropy estimation (MEE) methods for analyzing Heart Rate Variability (HRV), are described. Analysis results showed that low frequency components (LF:0.04-0.15 Hz) reflecting sympathetic nervous activity occupy a dominant position on day time, high frequency components (HF:0.4-0.40 Hz) reflecting vagal nervous activity become stronger at night. There is a cluster of peaks ranging from 0.20-0.30 Hz at night. In the morning, LF/HF ratio trend graphics changed sharply. This may be quite significant for predicting sudden death of an intensive heart disease patient. Both analysis results have a high degree of positive correlation, and cross-correlation coefficient is more than 0.99. However, MEE has a higher frequency resolution than FFT.